Infrared photography is an extraordinarily valuable technique for detecting preparatory underdrawings in paintings where the underdrawing is executed over a white priming with a medium containing carbon, such as ink, charcoal, or paint made with a carbon pigment—bone black, for example. Infrared radiation, invisible to the human eye, can penetrate a painting’s surface layers and capture an image with infrared-sensitive film or via specialized software on a computer screen.

Infrared photography and modified digital cameras work in the near-infrared waveband, with limited penetration (700–1000 nanometers). Pigments such as azurite and malachite block penetration at this wavelength, which is a major limitation since these pigments are used widely for robes, skies, and landscapes in medieval and Renaissance paintings. However, a technique known as infrared reflectography, developed by J. R. J. van Asperen de Boer in the late 1960s at the University of Utrecht, Netherlands, can penetrate much further, to 2500 nm. Unfortunately, the technique had low resolution and was labor-intensive. The arrival of solid-state digital technology in the mid-1990s transformed the technique. Experience with this technology shows that the waveband of 700 to 1700 nm provides an excellent image and can penetrate layers of azurite and malachite. Given that the waveband at which infrared images are taken will determine what might be revealed, it is important to know the waveband used so that accurate assessments and comparisons of underdrawings in art works can be made.

This was dramatically demonstrated in the examination of the Virgin and Child (fig. 1), from the workshop of Botticelli, in the Gallery’s collection. The painting is one of many versions of the composition, with variations in background details and revisions of the Christ Child’s pose. The composition is the centerpiece of the San Barnaba Altarpiece that Botticelli painted around 1484 and that is now in the Uffizi Gallery in Florence.

In 2004 an infrared image was taken of the Gallery’s painting with a modified digital camera that had limited penetration to about 1000 nm into the infrared spectrum (fig. 2). Some drawn outlines were already visible where the original paint had been mistakenly removed sometime before 1915, according to the Gallery’s records. The left hand of the Virgin and the Christ Child’s left leg had been cleaned down to the green underlayer, exposing the underdrawing outlining the Virgin’s hand. A few summary outlines could also be seen. The apparent absence of underdrawing elsewhere suggested the painting was a routine version made by Botticelli’s studio with little preparatory drawing. This perception was further strengthened by the work’s discolored varnish and by retouching that flattened and coarsened the image.  

In 2018 Laurence Kanter, Chief Curator and the Lionel Goldfrank III Curator of European Art, asked the Conservation Department to reassess the condition of the painting because he was convinced the work deserved further study despite its condition. It was decided that a full cleaning and limited restoration would enable a better judgment of the painting’s quality.

Cleaning dramatically improved the work’s appearance and uncovered the high quality of the painted heads of the Virgin and Christ Child, which are undoubtedly by Botticelli. The painting was reexamined with infrared reflectography, using an InGaAs camera (fitted with an indium gallium arsenide photodiode). This camera can penetrate into the mid-infrared spectrum, with a range of 700–1700 nm. The result was completely unexpected (fig. 3). A highly complex underdrawing emerged, revealing a wide range of preparatory drawing. Smooth brushed outlines appear to have been drawn around a template with lines at right angles in the bottom right corner that might have been drawn to position the template. Elegant brushed lines indicate a collar to the Virgin’s robe, which was not painted, and complex folds in her gown are suggested. Thinner freehand lines carefully delineate the features and the hair of both figures. Alterations to the position of the Virgin’s fingers and the Child’s feet are visible.

At the painting stage, the drawn outlines of the Virgin’s robe were not followed closely. A tower sketched on the left was not painted. The Virgin’s hair and folds of her head covering do not follow the detailed underdrawing. The faces of the Virgin and Child, however, do follow the underdrawing more closely.

There is still a great deal to be learned. It would be fascinating to compare versions of the composition in private collections and other museums using infrared reflectography at the same waveband. This might reveal much about the working practice of Botticelli’s large studio. Was there a stock model that was then modified by Botticelli or an assistant? Would an assistant have been allowed to make changes from the underdrawing when painting the image? And does the execution of the beautifully painted heads—with subtle improvements over the drawing, such as the more sinuous curve of the Virgin’s neck—suggest the master’s hand, while the simpler delineation of the Virgin’s robes indicates that ­someone other than Botticelli finished the painting?